In the present paper, a numerical study on the flow control over a backward facing step using dielectric barrier discharge (DBD) plasma actuators has been conducted by means of high-fidelity computational fluid dynamic and multi-objective design optimization (MDO) based on surrogate-assisted evolutionary algorithms. The main objectives of this study are minimizing total pressure loss and reattachment length while maximizing the flow uniformity index. It employs four decision variables including input voltage, frequency, width of generated plasma and distance from the flow inlet to the start of the generated plasma. Sensitivity analysis has been performed with the aid of surrogate modeling; the results have been evaluated through MDO with evolutionary algorithms. It has revealed major impact of the DBD plasma actuator on the behavior of the flow and major improvements on the objective functions in relation to the decision variables. In particular, flow separation has been suppressed considerably while maintaining reasonable levels of flow uniformity and total pressure loss.